A physicist's algorithm spots the patterns behind terrorist attacks
A physicist has developed an algorithm to spot the patterns behind terror attacks that he says could be helpful in predicting and even preventing future attacks.
Neil Johnson, a professor of physics at the University of Miami in Florida, assembled a team of physicists to create a mathematical model that analyzes pro-Islamic State posts on social media sites.
"We were really interested in the hardcore pro-ISIS support," he tells guest host Josh Bloch on CBC Radio's Day 6. "This is people who are exchanging operational details."
While intelligence agencies and law enforcement organizations monitor this kind of activity, Johnson says that his team's scientific approach looks at this data in an entirely different way.
It's not the individual. it's how they come together into groups.- Neil Johnson, physicist, University of Miami
How they do it
"We are looking at the collective level. It is not a piece of the water that is creating the boiling. The whole thing boils at the same time," Johnson says. "We have a very collective view that's an area of physics called complex physics."
Johnson's team's research was published last month in Science magazine and focuses on the Russian social media site Vkontakte.
His team looked at the way known ISIS sympathizers clustered into groups or "aggregates" on the social media site.
"It's not the individual. it's how they come together into groups," Johnson says. "The key is to understand the group ecology. That and the narrative that they discuss holds the key — we believe — to being able to say something about what's looming on the horizon."
Over a six-month period, his team tracked how these aggregates formed, disbanded, and converged in the lead up to terrorist attacks. Johnson says the rate at which these groups started to crystallize followed a very specific mathematical form.
They used this information to build a predictive algorithm which Johnson says could provide clues about the likelihood and timing of future attacks. But Johnson is careful to point out that while this research may be helpful to law enforcement, his model deals in probabilities only.